Why extra headroom is needed for MP3s

RE: Why extra headroom is needed for MP3s (Alegria)

Tue, 04 Jan 2011 15:32:08 GMT

Thanks for the tip, but I have my own site now (at a comparable cost to a VIP SoundClick account). And yes, I'm "pseudo-streaming" at 320 kbps. Theirs nothing like it.

RE: Why extra headroom is needed for MP3s (bitflipper)

Tue, 04 Jan 2011 15:15:13 GMT

Most hosting sites do stream at 128kb/s, ostensibly because they want the quality of paid downloads to be better by comparison. If you could stream at 256 or 320, you could then simply capture the stream and obtain the higher quality for free. Even Band Camp, which supports free downloads in any format, including uncompressed waves, streams at 128kb/s.

You're right, a VIP account only makes sense if you have ambitions of making money from downloads. Fortunately, I am unencumbered by any such designs and I'd like to offer high-quality free downloads. One way to accomplish this from a free SoundClick account is to link to another download site. Your ISP probably provides an adequate amount of free storage where you can copy your songs to, and sneak a link into your song descriptions on SoundClick.

RE: Why extra headroom is needed for MP3s (Alegria)

Tue, 04 Jan 2011 12:57:10 GMT

"bitflipper"

This applies to free Soundclick accounts only. Paid "premium" accounts just store the files you upload, without modification. But free accounts are restricted to 128kb/s CBR.

I had a VIP account at SoundClick for about 6 months, before I cancelled. One of the reasons I went for it in the first place was for the higher bit-rate mp3s. I was under the impression that they were going to be streamed at that rate. Not the case at all. They stream everything without exception @ 128 kbps. The higher bit-rate mp3 (unconverted) is only available for downloads. And although the information for this is hard to find on their site, "Tolgar" has confirmed this when directly asked about it.

In conclusion, a VIP account would make sense only if you're selling your works and want to offer the highest mp3 quality as a download only, after a purchase.

RE: Why extra headroom is needed for MP3s (patm300e)

Tue, 04 Jan 2011 12:48:19 GMT

Thanks. I am going to check out Audacity, but I think I might just use the LAME encoder called from inside X1a...

BTW, here is a link to the thread that contains info on LAME inside X1a:

http://forum.cakewalk.com/tm.aspx?m=2170770&high=LAME+Mp3+Setup

RE: Why extra headroom is needed for MP3s (bitflipper)

Tue, 04 Jan 2011 11:45:23 GMT

MediaCoder is still alive, but has been accused of being a carrier for adware.

Audacity might be an alternative.

RE: Why extra headroom is needed for MP3s (patm300e)

Tue, 04 Jan 2011 09:01:34 GMT

http://mediacoder.sourceforge.net/index.htm
Media Coder looks to be dead code now, any other good options?

RE: Why extra headroom is needed for MP3s (dontletmedrown)

Fri, 11 Dec 2009 18:47:39 GMT

mike_mccue

I figured I'd start with a "single" to see how it all works out.

It'll be 4-6 weeks before I expect to know if ITunes even accepts it.

Good god. I recommend using Tunecore instead of going direct. They have gotten some of my clients' onto iTunes in just a couple of days.

Re: RE: Why extra headroom is needed for MP3s (plectrumpusher)

Fri, 11 Dec 2009 17:20:39 GMT

bitflipper

I make sure the thingy doesn't go into the pink area, above the 6 in the track view. I should make the thingy go even lower?

"I make sure the thingy doesn't go into the pink area" - don't let those Coffee House jokers get hold of that one!

If you're working in a purely analog environment, the rule is simple: keep every meter in the signal chain out of the red, in order to avoid clipping at every stage. That applies to everything from your mic preamp to your effects to your tape machine. (OK, the tape machine is forgiving and is routinely hit hard, but even it has limits and can also produce unpleasant distortion when driven too hard.)

In the digital realm we're given a lot more room for sloppiness, at least in SONAR. That's because SONAR works with floating-point data internally. The handy thing about floating-point data is if the number ever gets too big, we just scootch the decimal point over and we're good. It is virtually impossible to damage a signal as long as it's stored and manipulated as floating-point data.

But digital-to-analog converters do not work with floating-point numbers. At that stage, our audio has to return to the world of integer, fixed-point representation. At that stage, you get 24 bits (or 16 bits) to work with, period. Once you've used all 24 (or 16) of them up then there's no place left to go, and severe distortion results.

But that nastiness only happens at the final stage of the recording process. Up until then, you can peg the meters into the red if you want to (it's still not a good idea, though) and no harm will come to your signal. (Yeh, there are esoteric arguments about loss of resolution, loss of bits, but don't worry about that - it's rarely a significant concern.)

What we're talking about here is the final output levels, which usually means the output of your mastering limiter. That's where you want to make sure you stay out of the red.

You also don't want to drive your limiter so hard that it has no headroom to work within. Drive the limiter too hot and it can't do its thing, which is to decide how much of that headroom to use. It will dutifully keep your output under control, but it may horribly mangle your sound in the process. Since most limiters have an input trim control, all you have to do is turn that control down. That way, no matter what you're throwing at the limiter it'll bring it down to a sensible level before applying limiting. If your limiter doesn't have an input control, you can use the bus trim instead.

So the short answer can be condensed to three rules:
1. Try to keep your tracks out of the red but don't obsess over it. 1 or 2db into the red is usually OK.
2. Make sure the levels going in to the limiter are -6db or lower
3. Make sure the output of the limiter is -1db to -3db, the latter recommended if the next step is MP3 encoding

Another option if you think you are square waving you transients is to create a null test and listen to the difference file that results , a real eye(ear !!) opener. Then just think of what the poor perceptual codec mp3 encoder is going to do with that !!

Re: RE: Why extra headroom is needed for MP3s (bitflipper)

Fri, 11 Dec 2009 17:06:25 GMT

I make sure the thingy doesn't go into the pink area, above the 6 in the track view. I should make the thingy go even lower?

"I make sure the thingy doesn't go into the pink area" - don't let those Coffee House jokers get hold of that one!

If you're working in a purely analog environment, the rule is simple: keep every meter in the signal chain out of the red, in order to avoid clipping at every stage. That applies to everything from your mic preamp to your effects to your tape machine. (OK, the tape machine is forgiving and is routinely hit hard, but even it has limits and can also produce unpleasant distortion when driven too hard.)

In the digital realm we're given a lot more room for sloppiness, at least in SONAR. That's because SONAR works with floating-point data internally. The handy thing about floating-point data is if the number ever gets too big, we just scootch the decimal point over and we're good. It is virtually impossible to damage a signal as long as it's stored and manipulated as floating-point data.

But digital-to-analog converters do not work with floating-point numbers. At that stage, our audio has to return to the world of integer, fixed-point representation. At that stage, you get 24 bits (or 16 bits) to work with, period. Once you've used all 24 (or 16) of them up then there's no place left to go, and severe distortion results.

But that nastiness only happens at the final stage of the recording process. Up until then, you can peg the meters into the red if you want to (it's still not a good idea, though) and no harm will come to your signal. (Yeh, there are esoteric arguments about loss of resolution, loss of bits, but don't worry about that - it's rarely a significant concern.)

What we're talking about here is the final output levels, which usually means the output of your mastering limiter. That's where you want to make sure you stay out of the red.

You also don't want to drive your limiter so hard that it has no headroom to work within. Drive the limiter too hot and it can't do its thing, which is to decide how much of that headroom to use. It will dutifully keep your output under control, but it may horribly mangle your sound in the process. Since most limiters have an input trim control, all you have to do is turn that control down. That way, no matter what you're throwing at the limiter it'll bring it down to a sensible level before applying limiting. If your limiter doesn't have an input control, you can use the bus trim instead.

So the short answer can be condensed to three rules:
1. Try to keep your tracks out of the red but don't obsess over it. 1 or 2db into the red is usually OK.
2. Make sure the levels going in to the limiter are -6db or lower
3. Make sure the output of the limiter is -1db to -3db, the latter recommended if the next step is MP3 encoding

Re: RE: Why extra headroom is needed for MP3s (brundlefly)

Fri, 11 Dec 2009 14:40:43 GMT

peggysuechan

I make sure the thingy doesn't go into the pink area, above the 6 in the track view. I should make the thingy go even lower?

It depends on whether the "thingy" is the RMS meter (red bar) or the peak meter (little red dot to the right of the bar, or by itself if RMS is not enabled). Keeping track peaks under -6dB and Master bus peaks under -1dB is a good rule of thumb. But RMS ("average") signal strength will typically be another 6dB or more below that peak level. If you don't have peak indicators enabled, and your RMS (the bar) is hitting -6dB, then the peaks could easily go over 0dB (clipping)... but you'd see that in the peak numbers and the peak warning marker at the end of the meter.

Re: RE: Why extra headroom is needed for MP3s (peggysuechan)

Thu, 10 Dec 2009 19:52:58 GMT

peggysuechan

I make sure the thingy doesn't go into the pink area, above the 6 in the track view. I should make the thingy go even lower?

Bump.

Re: RE: Why extra headroom is needed for MP3s (timidi)

Thu, 10 Dec 2009 19:10:35 GMT

thanks bitflipper. good info.

Re: RE: Why extra headroom is needed for MP3s (peggysuechan)

Thu, 10 Dec 2009 18:58:35 GMT

I make sure the thingy doesn't go into the pink area, above the 6 in the track view. I should make the thingy go even lower?

Re: RE: Why extra headroom is needed for MP3s (feedback50)

Thu, 10 Dec 2009 18:49:39 GMT

Check out the peak values on the main bus meter and adjust levels accordingly.

Re: RE: Why extra headroom is needed for MP3s (peggysuechan)

Thu, 10 Dec 2009 17:38:14 GMT

So how do we leave headroom? Who, what, where, when, why, and how in the world is this -1db that y'all're talking about?

RE: Why extra headroom is needed for MP3s (jcatena)

Sun, 14 Jun 2009 11:10:43 GMT

> I thought I said that in reponse to Dave's (bitflipper's) initial posts... Did I miss something? Too many "Dave's?"

Sorry Dave, you said it well. I only wanted to state two implications not mentioned before that you and many may know, but perhaps others don't. I was not aswering to you or anyone in particular.

RE: Why extra headroom is needed for MP3s (DaveClark)

Sat, 13 Jun 2009 12:34:08 GMT

Hi Jose,

Note that as Bitflipper wrote, not only mp3 is affected.

I thought I said that in reponse to Dave's (bitflipper's) initial posts... Did I miss something? Too many "Dave's?"

Regards,
Dave Clark

RE: Why extra headroom is needed for MP3s (jcatena)

Sat, 13 Jun 2009 10:20:08 GMT

Note that as Bitflipper wrote, not only mp3 is affected.
For example, any modern DAC does oversampling, where the interpolation can result in samples of higher values than any in the input. Furthermore, a steep antialiasing filter is applied, that will have some ripple. In most cases the filter itself applies some neagtive gain to reduce the probability of clipping, but you don't know how much margin it provides internally, if any. Any sample rate conversion also results in peak levels that can be higher than in the input. And most common DSP algorithms.
So definitely its good to leave some headroom, but how much? You can only guess depending on the distribution format. MP3 or any loosie format needs more, and the more as loosier.
Hopefully all players will use ReplayGain soon. If this happens it will solve many problems, the most important one that you can not make your song sound louder than others through more dynamic range compression, neither you need to destroy it in order to sound as loud as the worst one. So we could deliver our music with as much headroom and dynamics as we wish.
The music industry should have set a standard for levels, as the cinema industry did, but sadly this never happened, and today, at least in pop/rock/dance, everything is delivered so highly compressed that it is virtually impossible to master to similar average levels without seriously compromising the quality. Most stuff is well above -10 dB RMS, many even above -5, that's ridiculous and everything sound like crap. Even first line bands that used to wound well, sound now horrible. The solution is so easy, but nobody wants to be the first delivering music that requires to turn up the playback volume to sound as loud as other's. If finally ReplayGain becomes standard in most players, we couild finally do it the right way, and the user will not need to change the playback volume, will not notice anything but a better, undistorted sound.

RE: Why extra headroom is needed for MP3s (rob.pulman)

Fri, 05 Jun 2009 14:52:05 GMT

Thanks for that Bob. I just started a new thread about it, didnt see your answer here.

RE: Why extra headroom is needed for MP3s (bitflipper)

Fri, 05 Jun 2009 11:59:12 GMT

I get nothing but a hiss noise

That's usually the result of attempting to encode a 32-bit file. No standalone MP3 encoders I know of can handle floating-point files. Some can't even handle 24-bit integer files. You can, however, encode MP3 using full-featured editors such as Adobe Audition.

RE: Why extra headroom is needed for MP3s (Bob Oister)

Fri, 05 Jun 2009 04:46:56 GMT

Hi, Rob,

I switched to MediaCoder because my new DAW is running Vista 64, so Iâ€™m using the 64 bit version and I recently upgraded from version 0.6.2 to version 0.7.0 listed on their download page.

Maybe you didnâ€™t download the â€œAudio Editionâ€. They have four different editions with multiple versions on the download page. Go to the download page, click the link that says â€œMedia Coder 0.7.0 (active version), then on the next page click on the top choice, â€œMedia Coder Audio Editionâ€ on the next page select either 32-bit or 64-bit.

When the software is installed and running, I just click the â€œ+â€ button at the top left of the software to locate and select the Wav file that I want to convert to MP3, make sure the â€œLAME MP3â€ tab is selected on the right side of the middle section of the software, and then choose a quality preset from the dropdown box. When you choose any preset, you should be able to see the LAME command line in the box below change to reflect the command properties of whatever preset you choose. Next click the â€œStartâ€ button at the right of the top menu.

You should then see the fly-out encoding status box while itâ€™s working and then the â€œSuccessfulâ€ message when the new MP3 file is created.

The properties box at the top right shows the size of both the original Wav file and the new MP3.

Hope this helps!
Bob

RE: Why extra headroom is needed for MP3s (rob.pulman)

Thu, 04 Jun 2009 14:01:46 GMT

Bob I'm trying to convert wav to mp3 using the mediacoder at the moment. Everything looks ok, but when I go to play the final mp3 I get nothing but a hiss noise..no song lol.

Any ideas?

PS - I downloaded the 32 bit version mediacoder (not the 64 bit)...I'm using XP. Have I downloaded the right version of mediacoder?

Thanks

RE: Why extra headroom is needed for MP3s (Lanceindastudio)

Wed, 03 Jun 2009 05:41:59 GMT

Good stuff Bit- thanx Bro

RE: Why extra headroom is needed for MP3s (rm5700@optonline.net)

Tue, 02 Jun 2009 20:15:31 GMT

Great info, and interesting...thanks for this Bitflipper

RE: Why extra headroom is needed for MP3s (rob.pulman)

Tue, 02 Jun 2009 14:56:16 GMT

Thanks very much for the info.

Bob - I'll definitely check out the mediacoder, soon as I get home off this nightshift!

Cheers
Rob

RE: Why extra headroom is needed for MP3s (DaveClark)

Tue, 02 Jun 2009 09:30:12 GMT

Hi Rob,

The -b option sets the bit rate, for example

lame -h -b 320 file.wav file.mp3

gives very high quality.

Regards,
Dave Clark

RE: Why extra headroom is needed for MP3s (Bob Oister)

Tue, 02 Jun 2009 03:03:51 GMT

Hi, Rob,

Lately Iâ€™ve switched to using the free â€œMediaCoder Audio Editionâ€ as a front end for LAME. You can find it here: http://mediacoder.sourceforge.net/index.htm

Itâ€™s very user friendly and makes it easy to utilize all of the options available in LAME. You can easily switch conversion bitrates by choosing from a dropdown box full of choices.

Hope this helps!
Bob

RE: Why extra headroom is needed for MP3s (rob.pulman)

Tue, 02 Jun 2009 02:45:41 GMT

I know this is resurrecting an old thread, but what the hell.

Can anyone explain how I actually alter the bitrate for the LAME convertor? I looked at the link from Bitflipper, and I understand that by altering the '-Vn' number in the code (numbers from 0-9 I think they are), I'll achieve better conversion resulting in a bigger mp3 file.

Thing is, I can't actually figure out how to physically do it in LAME.

My mp3s now usually turn out about 3.5mb, so I could really have a better conversion rate and still not exceed the 10mb Soundclick limit.

Any help would be appreciated, I'm also thankful for the explanations given on this thread previously.

My wife says I've got a brain like a computer - information needs to be punched into it.

RE: Why extra headroom is needed for MP3s (7-string_guy)

Thu, 21 May 2009 20:11:28 GMT

thanks mr flipper. that was a great read

RE: Why extra headroom is needed for MP3s (No How)

Thu, 21 May 2009 13:12:22 GMT

What a blessing to find this thread.

these last couple weeks i've been rackin' my brain trying to figure why the mp3 is distorting when the wav isn't.

I keep going back into projects to take down more and more high end so it doens't screech... but to no avail.
of course an intelligent logical guy would immediately put 2 and 2 together.

I see i need to get something i've never gotten before....HEADROOM.
also: I'm using the mp3 converter in Audacity. I'll experiment with 'V'

(Thanks,Bit.)